Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


The Project That Wouldn’t End

If you were around during the bad old days of the first computerized music and digital audio tools, say 10 to 15 years ago, then perhaps you remember,

If you were around during the bad old days of the first computerized music and digital audio tools, say 10 to 15 years ago, then perhaps you remember, or perhaps you’ve mercifully forgotten, some “features” you were forced to deal with:

  • Loss and/or corruption of files.
  • Loss of sync at odd and unpredictable moments. Momentary dropouts and digital “clicks” at random times.
  • Programs that crashed just when you finished a complicated session, but before you saved.
  • Programs that wouldn’t save.
  • Programs that wouldn’t quit.
  • Clumsy modal programs that made you switch gears entirely when you went from assembling to editing, or vice versa.
  • Blind signal-processing functions, in which you had no idea what a process sounded like until after you executed it.
  • Processing functions that refused to do anything.
  • Processing functions with fixed, and useless, parameters.
  • Non-undoable functions.
  • Functions that took enormous amounts of time, but were uninterruptible.
  • File-compatibility issues among different programs.
  • File-compatibility issues among different versions of the same program.
  • Lousy technical support.
  • No technical support.
  • Undocumented features, and documented features that didn’t exist.
  • Documentation that’s just plain wrong.
  • And that perennial favorite: blanket denial on the part of the developers that there was anything amiss.

Aren’t you glad that we’ve gone beyond all of this — well, most of it? Well, don’t count your chickens quite yet, because there’s a whole new world that’s becoming more and more important in the work that a lot of us do. In this world, each of these problems is back and in force, and ready to bite us on the behind. I know, because thanks to these new tools, it has taken me nine months to get a project done that should have taken about six weeks. But before you panic, let me tell you I’m not talking about audio: I’m talking about video. Specifically, Digital Video, aka DV.

If you’re not familiar with DV (which I assume means you live on the Outer Hebrides without a satellite dish), it has taken over the low- and medium-budget video production markets even faster and more dramatically than digital audio workstations did in our part of the universe. Those $2,500 broadcast-quality camcorders, $8.00 60-minute tapes and editing decks the size of a Happy Meal have revolutionized the industry. Done right, the quality can be fantastic; I’m told that a producer at WGBH, my local PBS station, did some tests when it first came out and found such little visible difference between DV and their standard format, BetaSP, that he recommended their producers adopt it. Computer editing systems that use the format work on the actual digital video data, not pointers to spots on a tape, or temporarily converted or transcoded analog video. So there is no generation loss anywhere in the chain, and there’s no distinction between online and offline editing. Interfacing various devices — cameras, decks, editors, video cards — is trivial, thanks to the near-universal adoption of FireWire in the format.

There are various versions of DV (mini-DV, DVCAM, DigiBeta, etc.), but they are almost all plug-compatible, and the format can accommodate different resolutions, color spaces and aspect ratios, so in its higher-end incarnations, it’s a perfect medium for producing HDTV. Like 24-bit audio, you need the right converters and more storage space to handle the higher-resolution formats, but the technology is basically the same as the cheap stuff.

Whatever resolution you work at, you will need gobs and gobs of storage to do a project of any length, but now that we have $150 (and falling) 60-gigabyte (and growing) removable hard drives, that’s not really an issue. Perhaps best of all, the same way that software plug-ins have forever changed the way we process audio (even if more and more plug-ins are emulating hardware processors of yore), in the DV world, if a software engineer can dream up some way to modify a picture or a scene, then it can be done.


My friend and colleague Howard Woolf, a longtime photographer and filmmaker, has been very excited about DV for a while. He considers it the first video medium that comes anywhere near film in its quality and flexibility, and it is available at a very small fraction of the cost. Woolf works both in the administration and in the Multimedia Arts department at Tufts University, where he teaches courses in digital video production and is faculty adviser to the school’s closed-circuit TV station. He knows his stuff. In early 1998, he put together his first DV editing suite, using a Mac-based program called “Edit DV” from a company then known as Radius.

Radius, for a long time, was one of the most important makers of monitors and video cards for Macintosh computers (remember the Pivot?). Edit DV, in fact, began life as an A/D video editing program called RadiusEdit. In that incarnation, and in its initial DV-only form, it was available only on the Mac, but it didn’t take long for Radius to see the handwriting on the wall. “Since they were a Mac company, Mac development there was originally ahead of PC development,” says Woolf, “but they looked at the market, and by the time that Edit DV 2.0 came along some 18 months later, they had put PC development ahead. The Mac people were livid.

“There’s no question in my mind that the Macintosh architecture is better than Windows for DV,” he continues. “They’re designed for this kind of work, although the gap is closing, and Windows 2000 is a major improvement. Unfortunately, our initial Edit DV experience was with the first Macintosh G3s, which had just come out and had terrible bugs in their onboard video, which you couldn’t get away from, even if you were using a dedicated video card. When Radius brought out its PC version, we jumped, which set us on a path we’ve been forced to continue down.” Howard’s lab (in the room next to mine), with 10 Windows 2000 workstations, would today be all Macs, probably running Final Cut Pro, were it not for those bugs.

Other companies were trying to bring Adobe’s Premiere into the DV world, with new drivers, plug-ins and machine control, but Radius got there first. The big boys in nonlinear video editing, Avid and Media 100, were using a completely different storage format, a form of M-JPEG, and told DV users that they would have to transcode their signal in order to use their systems. “They tried to convince people that DV-to-DV transfers were somehow ‘missing something,’” says Woolf, “which, of course, was ridiculous, since it’s a direct data transfer.” It wasn’t until very recently that Avid came out with a DV system, but even now they don’t make their own hardware for it.

The first version of Edit DV was actually two programs, much like the first, awful version of Digidesign’s Pro Tools. “Moto DV” was the software that controlled the video decks and captured the video into the computer, while Edit DV was the editing program. Soon after Woolf had committed to the PC base, Radius — then known as Digital Origin — came out with Version 2.0 of Edit DV, which combined the two programs. It wasn’t a bad program, although it had some interesting glitches. Perhaps the worst was that the next version was very long to follow — almost exactly two years, which is two centuries in computer product-cycle time — and was even worse. In the meantime, Digital Origin had been bought by Media 100.


A year ago September, I was invited by the Acoustical Society of America to present a paper on a recent musicological project I was involved with at their convention in California that December. I was flattered that they asked, but with no one to pay my expenses, it didn’t seem like something I could do. I asked the committee whether I could submit a video instead. They said, “Sure,” and thus my fate was sealed.

I called up Howard and presented the idea to him, and he was very enthusiastic, lining up three students to help with the editing, and even using his connections to get some funding from the university provost so that I could get paid for writing the film and he could buy some new equipment.

I’m not a filmmaker, and I don’t even play one on TV, but I’ve done music and sound for dozens of documentary films, and I know more than a little bit about how they’re put together. So I was going to write the script, and, in my best friendly professor voice, narrate the film. Howard and the students were going to shoot me talking, log and edit a large amount of footage I already had relating to the project, and shoot new footage where necessary: interviews, location shots and quite a few stills. Howard and I were going to be co-producers, but he was going to make most of the editing decisions. When he had a rough cut done, we’d go through it together. I’d offer suggestions, and he would make a final cut. Sounded pretty straightforward.

Howard and his students stuck me in front of a blue screen and shot me and recorded my narration, and over the next couple of weeks logged and digitized all the footage, while he and I went back and forth over the script, cutting down my 45 minutes or so of copy into something that would fit the 20-minute slot allotted to me. But then, for some reason, I never got to see any of the edited footage. Howard kept assuring me that it was going to be fine, but every time I asked, it seemed it wasn’t quite ready for me to look at yet.

Now it was late November, and it was a week before we were supposed to send the finished video out to California, but I still hadn’t even seen a rough cut. One evening, I was scheduled to give a lecture to some music department students and faculty, during which I hoped to show a “preview” of the video. As I was about to start the lecture, and the tape hadn’t arrived yet, Don, one of Howard’s student assistants, came in breathlessly and said that the film would be late. “Something went wrong when we were printing it to tape,” he gasped, “and we have to do it over again.” Fortunately, I was lecturing on two different topics, so I just reversed my planned order. An hour later, Howard arrived with a videotape.

We put it in the projector, and what I saw was 12 minutes of jumpily edited material, with flash and “ghost” frames around the edits, audio that kept jumping up and down in level, not to mention side to side between channels, sync that went in and out, and stills full of vibrating black lines. It was pretty awful, and I was pretty scared.

What had happened? Well, as Howard put it, the system “hiccupped.” It happened during the rendering process, which is the step you have to take after you’re done editing, during which all of the crossfades, wipes, sound level changes, zooms, pans and other effects are saved in separate, reprocessed files that the manufacturer calls “generated media.” Once all those files are created, the program is printed to tape. “There’s a lot of room for real wacko things to happen during rendering and printing,” explained Howard. “If you followed the instructions from the company, you would put all your files — pointers, clips and rendered video — on the same C drive. But this goes against the oldest law on the books, which is that you keep those files on separate disks.” Sound familiar?

“If they’re all on one disk, when you want to print the program has to do three things at once. And even though we had dedicated disks for clips and rendered video, the longer and more complicated the movie, the more calls from the program to the drive bus. You can end up with a roadblock, everything coughs and then you lose data, including your clip files and the pointers. This is what happened during what we thought was innocuous ‘save’ after rendering. Everything froze, and when we rebooted and went back in to Edit DV, the clips acted like they were totally corrupted, even from backed-up versions. The funny thing was, that when we played the clips in Quicktime, outside of Edit DV, they were okay — which was of no help, of course.”

Howard managed to reassemble a version of the program in the space of an hour, leaving all of the effects out, so that we could show it that evening.

All during the next weekend, Howard and Don literally worked night and day to get the piece back to where it was before the printing disaster, starting the assembly process from scratch and constantly running into file-corruption fallout — and almost ruining their Thanksgiving in the process. On the afternoon we were supposed to ship the final cut to California, Howard burned three copies of the DV tape, and then made a VHS copy of one of them so I could watch it and finally see what I hoped was a reasonably finished cut.

I made the mistake of not looking at the tape until that night, and imagine my horror when I saw timecode numbers prominently displayed on the left half of the picture. One of the folks, harried to the point of distraction, forgot to shut off the Display switch on one of Howard’s cute little DV decks during dubbing, and it was generating visual timecode while it recorded. Howard couldn’t be sure whether the copy he sent to California had the burn-in or not, so late that night he ran off yet another copy, checked it to make sure that it was clean and sent it off in the morning, Same-Day Delivery to California. By this point, for what we were paying FedEx, I could have gone to the convention in person.

Next month: It ain’t over ’til it’s over.

Paul D. Lehrman is Web editor for Mix and its sister magazines, but that doesn’t mean he has to be nice about inferior entertainment media or crummy tools.